Jasonpandre wrote

When doing a speed test between AP/Cleints I often get errors (Date note found or similar).

 It's »No data found«. The AJAX call of the web UI didn't succeed. You get this error when starting the speed test client over a wireless connection on the remote CPE.

Solution is to start the speed test server on the remote CPE and then run the test client on the local (or on the »most local«) CPE. You can even log out from the remote's web UI, the speed test in server mode will continue to run.

When i can get a response it has been aroiund 61Mb. Signal for the clinets are both -101dB Noise strength,  32 to 37dB SNR.

IMO that's a pretty bad signal strength (64 to 69 dBm) for such short distances. SNR should be ~50 to 60dB, signal strength ~50 to 40dBm. Try to align the CPE's antennas more precisely to each other.

Set WiFi mode to 802.11n-only, not to mixed mode, MAXtream disabled, distance setting to »Auto«. I guess you use 20 MHz channel width? You could try 40 MHz, but note that 40 MHz allocates 8 channels (out of 11 or 13 in the 2.4 GHz band) at once, so interferences with nearby APs are most likely to occur if the 2.4 GHz band is overcrowded. That's why I prefer CPE510 for directional links.

But even with 20 MHz channel width you should be able to reach ~100Mbps half-duplex wireless throughput in 802.11n-only mode (max. possible WiFi rate - 30% overhead).

Is there a fundamental problem with daisy chaining the CPE210 in this way?

No, that's not a problem; it's a typical topology using a relay (the CPEs in the midst of the whole link form the relay), sometimes also called »hopple-hopple-hopple route« (each wireless link is a hop and the whole route can go around corners). This is the preferred topology if there is no free line of sight between the endpoints. But I would use static IP addresses for all CPEs and the EAP, not DHCP. DHCP is not suitable for stationary devices.

@Jasonpandre, of course that's fine to have DHCP for the mobile clients.

One aspect if you have the midst two CPEs on the same antenna pole is that they can slightly interfere with each other. I would mount them distant from each other, probably at different heights on the same pole, but preferably on their own poles if possible.

@Jasonpandre, there is something weird: CPE #1 uses channel 1, CPE #2 uses channel 11 to connect to CPE #1? Can't work this way.

Why do you use 10 MHz channel width? Any reason for this? 10 MHz limits max WiFi rate to 72 Mbps, so you cannot achieve more than 50 Mbps goodput (= throughput - WiFi protocol overhead).

Jasonpandre wrote

This was an extension of your idea re the 20Mhz.

I see. In the standards, channels are numbered at 5 MHz spacing within a band (except in the 60 GHz band), but transmitters usually occupy at least 20 MHz channel width.

eventually I solved the problem with a small movement of the final CPE unit. It was mounted on a low building with some small young trees in front, but not directly in line.

Ah, ok. Trees in between CPEs is not a free line of sight (or more precisely a required fresnel zone clearance). Trees will cause distortions of the WiFi signal in directional links.

Glad that it works now. Have fun with your CPEs!